I don't see any party here that needs rescuing by the government.
There are obviously holes in this narrative, but we like telling ourselves this can happen.
While there's not enough value in 17th century art history for the US to graduate 200k scholars per year and expect in-field employment, there is value to society in general to not losing our history and learning more. Letting the banks decide who gets loans will instantly mean those who can't pay cash cannot pursue that education or need to go on cheaper routes for college. They may or may not be viewed as a comparable education and quality candidate.
I'm not a fan of our current system, but a purely utilitarian system I see proposed seems likely to continue the spreading the social divides between the rich, middle class, and poor.
I'm an advocate for cutting a lot of the unnecessary fluff and expenses from college. Administrative overhead is excessive. Dorms are way nicer than they need to be. Alas, other colleges have nice dorms, so a single college choosing to go cheap on the dorms will probably decrease admissions and thereby revenue.
Drug abuse is less long-term priced in and likely also has higher aggregate risk as well.
In a sense the things you list are kind of like the broken windows fallacy. The increase in healthcare spending shows up in our GDP, but isn't an efficient use of the resources.
I also imagine that student loan debt could reduce local growth. If you have a graduate using most of their income on repaying loans, that's money that may have been spent locally and is instead sent to the bank. Of course the opposite is true before graduation, where inflated costs of education result in more money in the local economy. However due to interest it may be the case that the former outweighs the latter.
Suppose that someone invents a better version of ELIZA, and everyone in the psychotherapy business is suddenly out of work.
Or slightly less absurdly, suppose that Intuit and H&R Block mess up their next round of lobbying, and the IRS implements that one-click tax payment we've all been asking for. And the accountant industry shrinks by a few percent and wages go down for the ones that are left.
Either of those would be a significant shift in the job prospects of a large group of people, which could affect how many student loans actually get repaid.
I think it might also count as a risk if it suddenly became fashionable to get degrees in, say, underwater basket weaving. As in, incoming students aren't perfectly rational with perfect information about what degrees have what job prospects, and I'm assuming the college loan system might not react quickly if something caused an overall shift towards less economically useful degrees.
It does lack one "feature" that helps the bad mortgages cause problems, in that I don't think it would be self-reinforcing. From what I understand, subprime defaults were contagious in a way that I don't think this would be.
Contrary to public perception, there's no robust investment market for student debt. US government ends up purchasing most of it through Treasury. In theory it's a great deal, as it gets to borrow at low Treasury rates and originate at higher rates.
I don't have a link to the original source, but "In September 2018, the U.S. Treasury Department revealed in its annual report that student loans account for 36.8% of all U.S. government assets." https://www.investopedia.com/articles/personal-finance/08121...
Edit: here's a more updated analysis of the Treasury disclosures, which puts the figure at 51.2% as of September 2019 https://www.advisorperspectives.com/dshort/commentaries/2019...
Devaluing those assets (which Treasury marks to market) would significantly dampen (if not destroy) US government's borrowing ability as well as make some government entitlement programs [at least temporarily] insolvent.
Tuition for ETH Zurich around 800 francs per semester.